Shyam Nandan Rai

Shyam Nandan Rai is a researcher affiliated with Politecnico di Torino in Turin, Italy. His work focuses on advancing computer vision and deep learning techniques, particularly in semantic segmentation, anomaly detection, and generative adversarial networks. He has contributed to projects like Federated Learning frameworks and Mask Transformer-based anomaly detection systems. His research emphasizes robustness in challenging environmental conditions and open-world learning scenarios. Key contributions include methods for road scene analysis, atmospheric turbulence compensation, and few-shot self-supervised learning. Publications span top venues like ICCV and NeurIPS, with notable works on semantic segmentation compression and open-set recognition.

Research interests include:

• Generative Adversarial Networks (GANs)
• Open-Set Recognition and Anomaly Detection
• Federated Learning and Distributed Systems
• Self-Supervised Learning Techniques
• Environmental Robustness in Vision Models
• Online Learning and Adaptation

Notable projects include:

• Mask2Anomaly (ICCV 2023 Oral) for universal anomaly detection
• Federated Learning implementations in PyTorch
• Fluid: Few-shot deraining framework